Skip to content

v0.13.0.0
Compare
Choose a tag to compare
@kmantel kmantel released this 14 Jan 04:04
· 1293 commits to master since this release
v0.13.0.0
06f3006
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

[0.13.0.0] - 2023-01-13

Documentation

  • Remove invalid escape sequences (#2531)

MDF

  • OutputPort: include shape and type
  • Improve handling of more complex Functions
  • Set up stateful parameter index handling
  • Implement DriftDiffusionIntegrator

Parameters

  • Allow validate methods to reference other parameters (#2512)

Testing

  • Use the correct execution mode in test_transfer_mech_integration_rate_0_8 (#2471)
  • Tests, treewide: Add missing "composition" marks (#2472)
  • MDF: make onnx noise generation more flexible
  • Drop extra 'monitor' mechanism from the threshold modulation test
  • Use numpy instead of python math module
  • Use lambda function to get the name of tested function
  • Consolidate Buffer Function test
  • Use more than one iteration in test using stateful functions
  • Run more than one iteration in tests that use stateful Functions
  • Consolidate test of different input formats
  • Consolidate testing of different mechanism functions
  • Run more than one iteration in tests that use integrator mode
  • Use 'num_trials' instead of manually calling 'execute'
  • Overload 'benchmark' fixture with custom invocation
  • Use the benchmark fixture unconditionally
  • Use the benchmark fixture unconditionally
  • Use the benchmark fixture unconditionally
  • Use the benchmark fixture unconditionally
  • Use custom benchmark fixture (#2529)
  • Don't use the same position in default variable
  • Add derivative test when using OUTPUT_MDOE=ALL
  • Use keywords instead of hardcoded strings as parameters
  • Add type of output to names of SoftMax tests
  • Assert that func_mode is only one of the three valid options
  • Tests, functions/SoftMax: Don't use nested 'params' dir to set parameters
  • Use 'size' instead of 'content_size' in construction
  • Check for compiled variant sooner
  • Check for warning messages when there's a conflict between mech and function parameter
  • Drop 'DECISION_TIME' output port
  • Add custom name prefix to all components with default names generated during test collection
  • Use custom name prefix for Components instantiated during test enumeration (#2542)
  • Do not check for string value of numpy error embedded in MechanismError (#2569)
  • Use mac/windows results for numpy>=1.22 and AVX512 CPUs

TransferFunction

  • Add support for computing derivatives using base function output (#2538)

Function/SoftMax

  • Use correct dimensionality when calculating full Jacobian (#2539)

Functions/DriftDiffusionAnalytical

  • Re-associate floating point operations

Functions/ReLU,SoftMax

Functions/SoftMax

  • Restore correct computation of derivation
  • Add support for 2d inputs to SoftMax derivative (#2534)
  • Use "ALL" results to calculate derivative directly
  • Use 'squeezed' version of output when calculating Jacobian

Functions/Transfer

  • Consider 'bias' parameter in 0 check in ReLU derivative function

Functions/TransferFunction/Angle

  • Optimize python version

Functios/ReLU

  • Rename 'variable' arguemnt to 'input'

Github-actions

  • Add hash of requirements.txt to cache key of built pip packages (#2482)
  • Restrict scipy to <1.9.2 on x86 (#2502)
  • Use environment files to set outputs (#2506)
  • Restrict scikit-learn to <1.1.3 on x86 (#2511)
  • Fix setting of outputs
  • Consider 'doc_requirements.txt' when generating cache key in doc CI jobs
  • Consider 'dev_requirements.txt' when generating cache key in CI jobs
  • Fix detection of wheel cache directory location (#2513)
  • Restrict statsmodels to <0.13.3 on x86 (#2516)
  • Run only the latest instance of the main CI workflows (#2537)
  • Check env var to tag selected OSes to run on self-hosted runners (#2546)
  • Generalize package name extraction from dependabot branch name (#2566)

Github-actions/codeql

  • Skip CI job on pushes that only update documentation (#2560)

Github-actions/coedql

  • Run codeql analysis on PRs to devel

Github-actions/install-pnl

  • Only check version of a new package if it's in a requirements file
  • Restrict jupyter-server to <2 (#2555)

Github-actions/install-pnl/dependabot_pr

  • Cleanup installation of the bumped package before installing PNL (#2481)
  • Skip package version check for not installed packages (#2483)

Library/AutodiffComposition

  • Drop empty docstring (#2567)

Llvm

  • Add sin, cos builtins
  • Implement compiled Angle and SoftMax derivative Functions (#2528)

Llvm/builtins

  • Implement binomial distribution for N==1 (#2552)

Llvm/codegen

  • Use visit_Constant instead of visit_{Num,NamedConstant} if available

Maint

  • Ignore mdf test autogenerated files (#2527)

Models/Botvinnick

  • Use top level import

Registry

  • Add configurable prefix to auto generated component names

Requirements

  • Restrict modeci_mdf to 64bit cpython (#2473)
  • Update pytest requirement from <7.1.3 to <7.1.4 (#2478)
  • Update pandas requirement from <1.4.4 to <1.4.5 (#2477)
  • Update matplotlib requirement from <3.5.4 to <3.6.1 (#2486)
  • Update pandas requirement from <1.4.5 to <1.5.1 (#2488)
  • Update autograd requirement from <1.5 to <1.6 (#2493)
  • Update pytest-cov requirement from <3.0.1 to <4.0.1 (#2492)
  • Update pandas requirement from <1.5.1 to <1.5.2 (#2503)
  • Update matplotlib requirement from <3.6.1 to <3.6.2 (#2497)
  • Update torch requirement from >=1.8.0,<1.12.0 to >=1.8.0,<1.13.0 (#2504)
  • Update pytest-xdist requirement from <2.6.0 to <3.1.0 (#2510)
  • Update pytest-benchmark requirement from <3.4.2 to <4.0.1 (#2507)
  • Update pint requirement from <0.20.0 to <0.21.0 (#2509)
  • Update pillow requirement from <9.3.0 to <9.4.0 (#2515)
  • Update pytest requirement from <7.1.4 to <7.2.1 (#2508)
  • Update matplotlib requirement from <3.6.2 to <3.6.3 (#2525)
  • Update pandas requirement from <1.5.2 to <1.5.3 (#2548)
  • Update torch requirement from >=1.8.0,<1.13.0 to >=1.8.0,<1.14.0
  • Update torch requirement from >=1.8.0,<1.13.0 to >=1.8.0,<1.14.0 (#2553)
  • Update pytest-xdist requirement from <3.1.0 to <3.2.0 (#2554)
  • Use upper and lower bound for fastkde (#2561)
  • Update pillow requirement from <9.4.0 to <9.5.0 (#2565)
  • Bump fastkde from 1.0.19 to 1.0.20 (#2559)
  • Update networkx requirement from <2.9 to <3.1 (#2570)
  • Update toposort requirement from <1.8 to <1.9 (#2573)
  • Update toposort requirement from <1.9 to <1.10 (#2575)
  • Update matplotlib requirement from <3.6.3 to <3.6.4 (#2576)
  • Update numpy requirement to <1.22.5
  • Reassociate operations in the DriftDiffusionAnalytical function and update numpy to <1.22.5 (#2476)

Treewide

  • Drop mentions of ControlMechanismRegistry
  • Drop mentions of GatingMechanismRegistry
  • Cleanup. Reduce the number of warnings issued during test execution (#2541)

Utilities

  • Check that the compared value is not an empty list
  • Add fastpath when checking if object is compatible with itself
  • Safe_equals: handle dict-like objects in np arrays (#2551)
Assets 3
Loading